INFORMS 2021 Program Book

INFORMS Anaheim 2021

WC34

2 - Fair Dynamic Rationing Vahideh Manshadi, Yale University, Quincy, MA, 02169-4688, United States Our paper studies a fundamental sequential resource allocation problem in which we aim to equitably and efficiently ration a divisible good to a sequence of agents with stochastic correlated demands. Focusing on the objective of maximizing the minimum fill rate, we show that an adaptive policy which is both simple and transparent simultaneously achieves the best possible ex-ante and ex-post guarantees. Our work is motivated by rationing social goods in situations where demands (needs) realize sequentially and are possibly correlated, such as a pandemic or a hurricane impacting multiple localities. 3 - Revenue Management for Boutique Hotels: Resources with Unit Capacities and Itineraries over Intervals of Resources Mika Sumida, University of Southern California, Los Angeles, CA, 10128-5805, United States, Paat Rusmevichientong, Huseyin Topaloglu, Yicheng Bai We consider the revenue management problem for a boutique hotel offering unique rooms. Customers arriving into the system make booking requests for different intervals of stay. The goal is to find a policy that determines an assortment of rooms to offer to each customer to maximize the total expected revenue. Because each room is unique and customers book intervals of days, the problem has two special features. First, each resource has a unit capacity. Second, the resources can be ordered such that each itinerary consumes an interval of resources. We develop two policies and give performance guarantees for both policies. We also develop an approach for computing an upper bound on the optimal total expected revenue. We give computational experiments on both a real-world boutique hotel dataset and synthetic datasets. WC32 CC Room 208B In Person: Pricing Innovations General Session Chair: Aydin Alptekinoglu, Pennsylvania State University, University Park, PA, 16802-3603, United States Co-Chair: A. Serdar Simsek, University of Texas-Dallas, Richardson, TX, 75080-3021, United States 1 - Pricing With Samples Amine Allouah, Facebook, Menlo Park, CA, 10027, United States, Achraf Bahamou, Omar Besbes we study a fundamental data-driven pricing problem in low information environments: how should a decision-maker (optimally) price based on a finite and limited number of samples from the distribution of values of customers. The decision-maker’s objective is to select a general pricing policy with maximum competitive ratio when the value distribution is only known to belong to some general non-parametric class. We study achievable performance for two central classes: regular and monotone hazard rate (mhr) distributions. At a higher level, this work also provides insights on the value of samples for pricing purposes. For example, against mhr distributions, a single sample guarantees $64\%$ of the performance an oracle with knowledge of the distribution would achieve, two samples suffice to ensure $71\%$, and ten samples guarantee $80\%$ of such performance. 2 - The Winner’s Curse in Dynamic Forecasting of Auction Data: Empirical Evidence from Ebay A. Serdar Simsek, The University of Texas at Dallas, Naveen Jindal School Of Man., Richardson, TX, 75080-3021, United States, Ernan Haruvy, Meisam Hejazi Nia, Özalp Özer Dynamic auctions typically have a stage of high exploratory activity, followed by an inactivity period, and then a final stage of last-minute bids with sharp jumps. In addition, when bidder valuations are correlated, bidders decrease their bids to account for winner’s curse. We develop a methodology to predict final stage bids in an auction and quantify the impact of winner’s curse. Our approach improves the forecast of a sample of eBay auctions’ final stage bids by 13.6%, on average, compared to a commonly used benchmark. We find that bidders in common value auctions decrease their bids by 29% due to the winner’s curse adjustment. Auction platform managers can use our methodology and results to support their choice of auction-specific minimum bid increments. Platforms can also decide how to allocate their resources to mitigate the adverse effects of winner’s curse adjustment.

WC34 CC Room 209B In Person: Transportation-Operations II Contributed Session Chair: Jose H. Ablanedo-Rosas, University of Texas at El Paso, El Paso, TX, 79912, United States 1 - Transit Network Design with Passenger Assignment Constraints Pramesh Kumar, University of Minnesota-Civil Egineering, Minneapolis, MN, United States, Alireza Khani The research proposes a bi-level optimization model for designing an efficient transit network. It considers both passengers’ and operator’s perspectives when deciding where to locate transit routes and determining their optimal frequency. It captures user behavior through the optimal strategy transit assignment model at the lower level. The overall model is a mixed-integer non-linear program that is solved using Benders decomposition. 2 - Designing a MATSim Environment for a One-way Car Sharing System as a Transport Mode Selin Ataç, EPFL, Lausanne, Switzerland, Nikola Obrenovi, Michel Bierlaire Car sharing (CS) services have become popular due to their financial and environmental benefits. The CS operators have offered flexibility by allowing one- way trips which resulted in vehicle imbalance in the service area. They have then introduced rebalancing operations to reduce the imbalance thus, to increase the level of service. Since it is exhausting to collect the data to develop a demand model, this work makes use of the agent-based simulation MATSim in a one-way CS system. The results are used to explain the relation between the city structure, demand structure and the different rebalancing strategies. 3 - Quantifying Shareability Potential in a Transportation Network: Introducing the Maximum Network Flow Overlap Problem Navjyoth Sarma, University of California-Irvine, Irvine, CA, United States, Michael Hyland The structure of transport networks and spatial distribution of demand vary across and within cities, affecting the viability of shared mobility modes ranging from fixed-route transit to on-demand ride-pooling. To inform the viability of shared mobility modes, this study proposes a modeling framework to quantify shareability potential within a city. The study introduces the ‘flow overlap’ metric to measure the number of trips sharing a path between an O-D pair. This concept is used to formulate the Maximum Network Flow Overlap Problem to assign person flows to paths that maximize shareability. The model output provides unique insights to support design of multi-modal shared mobility systems. 4 - Traffic Equilibrium with Shared Mobility Services in a Coupled Morning-evening Commute Framework Wei Gu, University of Southern California, Los Angeles, CA, United States, Maged Dessouky, Jong-Shi Pang, Michael Zhang In this study, we develop a general equilibrium model to capture the complex interactions between solo-driving, rideshare and e-hailing that allows travelers to switch between different transportation modes in a coupled morning-evening commute. The model is formulated as a mixed complementarity problem. Then the existence of an equilibrium solution and the properties of the solution are investigated, and we provide conditions on the model parameters under which the equilibrium will be unique. The proposed model is validated with the renowned Sioux-Falls network. 5 - Operational Efficiency of Major Cargo Seaports in Some Developing Regions: Results from a Bootstrap DEA Jose H. Ablanedo-Rosas, University of Texas at El Paso, El Paso, TX, United States This study estimated the operational efficiency of 40 seaports located in five different developing regions. The study considered panel data from 2013 to 2018. The results showed that most of the ports were not operating at their optimal scale and reported an increasing and decreasing trend in their bias-corrected CRS and VRS efficiency scores respectively. One major finding is that most of the ports have experienced an increase in their operations without attaining the corresponding increase in their technical capabilities. The overall and regional top port performers were identified for benchmarking purposes.

165